Analyses the output from other functions within the packages to generate a basic descriptive summary for each accelerometer data file. Analyses include: Average acceleration per day, per measurement, L5M5 analyses (assessment of the five hours with lowest acceleration and with highest acceleration). Further, the traditionally popular variable MVPA is automatically extracted in six variants: without bout criteria in combination with epoch = epoch length as defined in g.getmeta (first value of the input argument windowsizes), 1 minute, and 5 minutes, and for bout durations 1 minute, 5 minutes or 10 minutes in combination with the epoch length as defined in g.getmeta.
g.analyse(I, C, M, IMP, qlevels = c(), qwindow = c(0, 24),
quantiletype = 7, L5M5window = c(0, 24), M5L5res = 10,
includedaycrit = 16, ilevels = c(),
winhr = 5, idloc = 1,snloc=1,mvpathreshold = c(),
boutcriter=c(),mvpadur=c(1,5,10),
selectdaysfile=c(),window.summary.size=10,
dayborder=0,bout.metric = 1,
closedbout=FALSE,desiredtz = c(),
IVIS_windowsize_minutes = 60,
IVIS_epochsize_seconds = 3600, iglevels = c())the output from function g.inspectfile
the output from function g.calibrate
the output from function g.getmeta
the output from function g.impute
array of percentiles for which value needs to be extracted. These need to be expressed as a fraction of 1, e.g. c(0.1, 0.5, 0.75). There is no limit to the number of percentiles. If left empty then percentiles will not be extracted. Distribution will be derived from short epoch metric data, see g.getmeta.
To specify windows over which all variables are calculated. If value = c(0,24) all variables will only be calculated over the full 24 hours in a day, If value =c(8,24) variables will be calculated over the window 0-8, 8-24 and 0-24. Previously this functionality was limited to the distribution in acceleration metric values, but now it also derives N valid hours, L5M5 analysis and MVPA.
type of quantile function to use (default recommended). For details, see quantile function in STATS package
Argument depricated after version 1.5-24. This argument used to define the start and end time, in 24 hour clock hours, over which L5M5 needs to be calculated. Now this is done with argument qwindow.
resoltion of L5 and M5 analysis in minutes (default: 10 minutes)
minimum required number of valid hours in day specific analysis (NOTE: there is no minimum required number of hours per day in the summary of an entire measurement, every available hour is used to make the best possible inference on average metric value per average day)
Levels for acceleration value frequency distribution in mg, e.g. c(0,100,200) There is no constriction to the number of levels.
window size in hours of L5 and M5 analysis (dedault = 5 hours)
If value = 1 (default) the code assumes that ID number is stored in the obvious header field. If value = 2 the code uses the character string preceding the character '_' in the filename as the ID number
If value = 1 (default) the code assumes that device serial number is stored in the obvious header field. If value = 2 the code uses the character string between the first and second character '_' in the filename as the serial number
Threshold for MVPA estimation. This can be a single number or an array of numbers, e.g. c(100,120). In the later case the code will estimate MVPA seperately for each threshold. If this variable is left blank c() then MVPA is not estimated
The variable boutcriter is a number between 0 and 1 and defines what fraction of a bout needs to be above the mvpathreshold
default = c(1,5,10). Three bout duration for which MVPA will be calculated
Functionality designed for the London Centre of Longidutinal studies. Csv file holding the relation between device serial numbers and measurement days of interest.
Hour at which days start and end (default = 0), value = 4 would mean 4am
Functionality designed for the London Centre of Longidutinal studies. Size in minutes of the summary window
This argument used to be called mvpa.2014 and had TRUE or FALSE as its value. However, it has now become clear that this aspect of the analyses is still very much open for debate. Therefore, I have changed it into an argument where you can specify a metric for bout detection based on a number. A description of these bout metrics can be found in the new function g.getbout
If TRUE then count breaks in a bout towards the bout duration. If FALSE then only count time spent above the threshold towards the bout duration.
see g.getmeta
Window size of the Intradaily Variability (IV) and Interdaily Stability (IS) metrics in minutes
Epoch size of the Intradaily Variability (IV) and Interdaily Stability (IS) metrics in seconds
Levels for acceleration value frequency distribution in mg used for intensity gradient calculation (according to the method by Rowlands 2018). By default this is argument is empty and the intensity gradient calculation is not done. The user can either provide a single value (any) to make the intensity gradient use the bins c(seq(0,4000,by=25),8000) or the user could specify their own distribution. There is no constriction to the number of levels.
summarysummary for the file that was analysed (see details)
daysummarysummary per day for the file that was analysed (see details)
The value summary is a dataframe and comes with the following variables:
ID Participant id extracted from file header
device_sn Device serial number extracted from file header
dodylocation Body location extracted from file header
filename Name of the accelerometer file
start_time Timestamp when experiment started
startday Name of day when experiment started
samplefreqSample frequency (Hz)
device Name of the device brand, e.g. Geneactiv
clipping_score Fraction of 15 minute windows per file for which the
acceleration in one of the three axis was close to the maximum for at least 80
percent of the time. This should be 0
meas_dur_dys Measurement duration (days)
complete_24hrcycle Fraction of 15 minute windows per 24 hours for which
valid data is available at any day of the measurement
meas_dur_def_proto_day Measurement duration
(days) minus the hours that are ignored at the beginning and end of the measurement
motived by protocol design
wear_dur_def_proto_day Measurement duration according
to protocol (days) minus invalid time periods
calib_err Estimated based on all non-movement periods in the measurement
after applying the autocalibration
calib_status Summary statement about the status of the calibration
error minimisation
ENMO_fullRecordingMean ENMO is the main summary measure of acceleration.
The value presented is the average ENMO over all the available data normalised per 24 hour
cycles (diurnal balanced), with invalid data imputed by the average at similar
time points on different days of the week. In addition to ENMO it is possible to
extract other acceleration metrics (i.e. BFEN, HFEN, HFENplus). We emphasize
that it is calculated over the full recording because the alternative is that
a variable is only calculated overmeasurement days with sufficient valid
hours of data.
pX_ENMO_mg_0-24h_fullRecording This variable represents the Xth
percentile in the distribution of short epoch acceleration values of the average
day within the time interval as specified.
L5hr_ENMO_mg_0-24_fullRecording Starting time in hours of the
least active five* hours within the time interval as specified (* window size defined by
argument winhr)
L5_ENMO_mg Average acceleration over L5
M5hr_ENMO_mg_0-24_fullRecording Starting time in hours of the most active
five* hours in the day within the time interval as specified (* window size defined by
argument winhr)
modifiable in g.getmeta)
M5_ENMO_mg_0-24_fullRecording Average acceleration over M5
ig_gradient_ENMO_0-24hr_fullRecording Intensity gradient calculated over
the full recording.
1am-6am_ENMO_mg_fullRecording Average acceleration between
1am and 6am
N valid WEdays Number of valid weekend days
N valid WDdays Number of valid week days
IS_interdailystability Intra daily variability
IV_intradailyvariability Intra intradailyvariability
AD_... The
variable ... was calculated per day and then averaged over all the available days
WE_...
The variable ... was calculated per day and then averaged over weekend days only
WD_...
The variable ... was calculated per day and then averaged over week days only
WWE_... The variable ... was calculated
per day and then averaged over weekend days. Double weekend days are averaged
This is only relevant for experiments that last for more than seven days
WWD_... The variable ... was calculated per
day and then averaged over week days. Double weekend days were averaged. This is
only relevant for experiments that last for more than seven days)
..._MVPA_E5S_B1M80_T100 MVPA calculated based on 5 second epoch setting
bout duration 1 Minute and inclusion criterion of more than 80 percent. This is
only done for metric ENMO at the moment, and only if mvpathreshold is not left blank
..._ENMO_mg... ENMO or other metric was first calcualte per day and
then average according to AD, WD, WWE, WWD
data exclusion strategy A log of the decision made when calling g.impute:
value=1 mean ignore specific hours; value=2 mean ignore all data before the first
midnight and after the last midnight
n hours ignored at the start of the measurement (if strategy = 1)
A log of the decision made when calling g.impute
n hours ignored at the end of the measurement (if strategy = 1)
A log of the decision made when calling g.impute
n days of measurement after which data is ignored
(if strategy = 1)
A log of the decision made when calling g.impute
The value daysummary is a dataframe and comes with the following variables:
id Participant id extracted from file header
filename File name
calender_date Calender data
bodylocation Body location (if known)
N valid hours Number of hours with valid data
N hours Number of hours of measurement
weekdayDay of the week
measurementday Day number relative to start of the measurement
L5hr_ENMO_mg_0-24h Starting hour of L5 on a scale from 0 to 24, where
14.5 means 14:30. Within the time window as specified
L5_ENMO_mg_0-24h Magnitude of average acceleration during the least
active five hours calculated with metric ENMO. Within the time window as specified
M5hr_ENMO_mg_0-24h Starting hour of M5 on a scale from 0 to 24, where
14.5 means 14:30. Within the time window as specified
M5_ENMO_mg_0-24h Magnitude of average acceleration during the most
active five hours calculated with metric ENMO. Within the time window as specified
mean_ENMO_mg_1-6am Mean acceleration between 1am and 6am
mean_ENMO_mg_0-24hr Mean acceleration over 24 hour period
pX_ENMO_mg_0-24h Percentile in the short epoch distribution with
invalid data imputed. Within the time window as specified
[A,B)_ENMO_mg_0-24h Time spent in minutes between (and including)
acceleration value A in mg and (excluding) acceleration value B in mg. This is
only done for metric ENMO at the moment, and only done if ilevels is not left blank
MVPA_E5S_B1M80_T100_0-24hr MVPA calculated based on 5 second epoch
setting bout duration 1 Minute and inclusion criterion of more than 80 percent.
This is only done for metric ENMO at the moment, and only if mvpathreshold is
not left blank
ig_gradient_ENMO_0-24hr Gradient from intensity gradient
analysis (Rowlands et al 2018) based on metric ENMO for the time segment
0 to 24 hours
ig_intercept_ENMO_0-24hr Intercept from intensity gradient
analysis (Rowlands et al 2018) based on metric ENMO for the time segment
0 to 24 hours
ig_rsquared_ENMO_0-24hr r squared from intensity gradient
analysis (Rowlands et al 2018) based on metric ENMO for the time segment
0 to 24 hours
# NOT RUN {
data(data.getmeta)
data(data.inspectfile)
data(data.calibrate)
# }
# NOT RUN {
#inspect file:
I = g.inspectfile(datafile)
#autocalibration:
C = g.calibrate(datafile)
#get meta-data:
M = g.getmeta(datafile, desiredtz = "Europe/London",
windowsizes = c(5, 900, 3600),
daylimit = FALSE, offset = c(0, 0, 0),
scale = c(1, 1, 1), tempoffset = c(0, 0, 0))
# }
# NOT RUN {
#impute meta-data:
IMP = g.impute(M = data.getmeta, I = data.inspectfile)
#analyse and produce summary:
A = g.analyse(I = data.inspectfile, C = data.calibrate,
M = data.getmeta, IMP)
# }
Run the code above in your browser using DataLab